Daniel sent us this one — and it's a good one. He's been watching those Kennedy Steve ATC compilations on YouTube, the ground controller at JFK who sounds like he's mainlining espresso and fury in equal measure. Rapid-fire New York accent, zero patience, and some truly spectacular meltdowns when foreign pilots can't follow his instructions. But Daniel noticed something interesting — Kennedy Steve's most viral, most stressful exchanges are almost exclusively with aircraft on the ground. Not takeoffs, not landings, not departure. Taxiway instructions and ramp coordination. Which raises the question: why is coordinating aircraft on the ground paradoxically harder than keeping them apart in the air?
This is the thing that most people miss about those Kennedy Steve clips. They're funny, sure — you laugh at the pilot who can't find the cargo ramp, you laugh at the guy getting berated in a thick New York accent. But underneath the entertainment is something genuinely revealing about aviation's hidden bottleneck. Ground control is the pressure cooker nobody talks about.
Because from the outside, it looks like the easy part. You're just... While the real skill is up there at thirty-five thousand feet.
Right, and that assumption is exactly wrong. Stephen Abraham — that's Kennedy Steve's real name — has been a ground controller at JFK since the nineteen eighties. He's worked that same position for decades. And the reason his most famous moments are all ground control isn't personality, it's structural. The geometry of JFK creates a cognitive load problem that makes approach and departure look almost relaxing by comparison.
Let's start with the map. What does JFK actually look like from the tower?
Four runways — four L Right, four R, three one L, three one R — and over thirty named taxiways with more than a hundred intersections. The airport diagram looks like a plate of spaghetti that someone dropped on a grid and then rearranged while angry. And pilots are looking at this on a moving map display that's often six inches wide, mounted in the cockpit, while also managing checklists and radio calls and monitoring other traffic.
A hundred intersections. That's not an airport, that's a maze with jet bridges. And I'm trying to picture this — you're a pilot who's never been to JFK before. You've studied the chart, sure. But studying a diagram in a quiet hotel room the night before is very different from interpreting it in real time while rolling down a taxiway at night in the rain.
Night and rain — that's the killer combination. Because here's what the chart doesn't tell you. At night, taxiway signs are illuminated but they're small, and if you're in a wide-body aircraft, your cockpit is three stories off the ground. The angle at which you're looking down at those signs through a rain-streaked windshield means they're often partially obscured. You're trying to read letters that are maybe eighteen inches tall from a height of thirty feet, at an oblique angle, while moving. And the signs are mounted at ground level. It's like trying to read street signs from a third-floor window while driving.
The chart is a clean, idealized representation, and the reality is a blurry, dark, wet mess.
And this is where the gap between theory and practice becomes a chasm. I've talked to pilots who fly into JFK regularly, and even they say that certain intersections — particularly where taxiways Alpha and Bravo converge near the approach end of four L — are confusing even in broad daylight. At night, in weather, after a long flight, they're disorienting. And that's for pilots who speak English natively and have been to JFK dozens of times.
Let's talk about that disorientation, because I think it connects directly to what Daniel was asking about. He mentioned head-up displays and detailed charting. Why don't those solve the problem?
They help, but they introduce their own issue, which is what I'd call the attention-splitting problem. A head-up display projects information onto a transparent screen in front of the pilot so they don't have to look down. Great for landing. For taxiing, it's less useful because the information you need — taxiway labels, hold-short lines, other aircraft — is on the ground, not in the air. You're looking down through the windshield, not out at the horizon. So the HUD is actually displaying information in the wrong focal plane. Your eyes are focused on the pavement twenty feet ahead, and the HUD is focused at infinity. It's a mismatch.
I hadn't thought about the focal plane issue. So the technology that works beautifully for approach and landing is almost working against you on the ground.
It's not actively harmful, but it's not the solution either. And the moving map display — the little screen that shows your aircraft superimposed on the airport diagram — that's better, but it has its own limitations. The GPS accuracy on the ground is typically within about ten to fifteen meters. That sounds precise, but when you've got two parallel taxiways that are sixty meters apart, and you're at an intersection where five taxiways converge, a ten-meter error can mean the difference between correctly identifying your position and being one taxiway off. And being one taxiway off at JFK is not a minor error. It means you might turn onto an active runway instead of a parallel taxiway.
Which brings us back to the margin-for-error point. In the air, a ten-meter position error is nothing. On the ground, it's potentially catastrophic.
Here's the asymmetry that nobody appreciates. In the air, aircraft are separated by miles horizontally and thousands of feet vertically. An approach controller at New York TRACON is working airplanes that are five miles apart, minimum, and often much more. You've got time. You've got altitude layers. If someone's a little slow on the readback, you've got thirty seconds before anything bad happens. On the ground at JFK, aircraft are separated by feet and paint. A two-second lapse in taxiway readback means a wingtip collision at fifteen knots. A wrong turn means an incursion onto an active runway. The margin for error goes from generous to zero.
It's not just that ground control is hard — it's that the consequences of a mistake are immediate and catastrophic in a way that approach control just doesn't face. And I want to sit with that for a second, because I think this is the core of Daniel's question. He's asking why Kennedy Steve sounds so stressed, and the answer is that the stakes are simply higher on a second-by-second basis than almost anywhere else in aviation.
Let me give you a concrete example that illustrates this. In twenty seventeen, there was an incident at JFK — not involving Kennedy Steve, but it's the kind of thing he's constantly trying to prevent. An EgyptAir 777 was taxiing to its gate and made a wrong turn onto a taxiway that was too narrow for its wingspan. The wingtip struck a parked aircraft. No one was hurt, but the damage was in the millions. The entire sequence happened in about eight seconds. The controller gave an instruction, the pilot read it back correctly, and then made a different turn than the one they'd acknowledged. By the time the controller saw what was happening and keyed the mic, the collision had already occurred. That's the window. In the air, eight seconds is an eternity. On the ground, it's the difference between a normal taxi and a multi-million-dollar insurance claim.
That's for a pilot who read back the correct instruction. So even when the communication loop appears to close properly, the execution can still fail.
Which is exactly what keeps ground controllers up at night. You can do everything right on the radio and still watch an aircraft turn the wrong direction. And this connects to something Daniel mentioned in his prompt — he wondered whether head-up displays and detailed charting have made taxi navigation reasonably easy. The answer is no, and the reason is what I'd call the no-autopilot factor. In the air, autopilot handles about ninety percent of the flight. The pilot is monitoring systems, talking to ATC, but the aircraft is essentially flying itself for long stretches. On the ground, it's one hundred percent manual. You're head-down in the cockpit, looking at charts, cross-checking taxiway signs through a windshield, while rolling at twenty knots. You're steering a hundred-and-fifty-ton aircraft with a tiller by your knee. There is no autopilot for taxi. There's no automation that says "turn left at Mike, hold short of four L." It's all human.
Which makes the language problem even more acute. Because Kennedy Steve's famous rants — the ones that go viral — they're almost always directed at foreign pilots whose English is shaky. And it's easy to watch those clips and think, well, these guys just need to learn better English. But that's not actually the issue.
No, it's not. The issue is that taxi instructions are the most linguistically dense communication in all of aviation. Let me give you a concrete comparison. A typical landing clearance is five to ten words. "Delta twelve thirty-four, cleared to land runway four right, wind zero four zero at twelve.A typical taxi clearance from JFK ground can be forty to sixty words with eight to twelve distinct instructions. I pulled one from a Kennedy Steve recording — it went something like: "JetBlue one two three, turn left on Alpha, right on Bravo, hold short of Mike, cross runway three one left at Mike, then left on Charlie, right on Echo, hold short of four L, contact ramp on one two niner decimal eight five." That's one transmission. One readback required. And if the pilot misses one turn, one hold-short point, one runway crossing — you've got a problem.
Forty to sixty words in a single burst, in a thick New York accent, delivered at the speed of someone who's been doing this since Reagan was president. And you're supposed to read it back correctly while also steering a jet. But I want to drill into something you said — that it's linguistically dense. What does that actually mean in practice for a non-native speaker?
Let's break it down. The instruction contains three distinct types of information layered on top of each other. First, spatial navigation: left on Alpha, right on Bravo, left on Charlie, right on Echo. That's a sequence of turns you have to hold in working memory. Second, procedural constraints: hold short of Mike, cross runway three one left, hold short of four L. Those are regulatory instructions — some of them are mandatory, some are conditional, and the consequences of missing one are not the same. Crossing a runway without clearance is a completely different category of error than missing a hold-short point on a non-movement area. Third, administrative instructions: contact ramp on this frequency. That's a frequency change that happens at a specific point in the sequence. So you're holding spatial, procedural, and administrative information in your head simultaneously, in a second language, while physically operating the aircraft.
The linguistic structure of English makes this worse, doesn't it? Because the prepositions — "hold short of," "cross at," "turn left on" — those are the small words that non-native speakers often process last or miss entirely.
The prepositions carry the critical safety information. "Hold short of Mike" and "cross Mike" are completely different instructions, and the only difference is the preposition and the word "short." If you're a non-native speaker under cognitive load, your brain is going to latch onto the content words — "Mike," "runway," "left," "right" — and the function words that specify the relationship might get lost. And Kennedy Steve's delivery, which is optimized for speed, doesn't give you time to mentally reconstruct the grammatical structure. You have to process it in real time, as it's coming at you.
His efficiency, which is his greatest asset, is also what makes him hardest to understand for the pilots who need the most clarity.
It's the central tension of his style. He's optimized for the system, not for the individual pilot. And in a system with no slack, that optimization makes sense — until it doesn't. Until you get a pilot who's on their first JFK arrival, who's been awake for eighteen hours, whose English is functional but not fluent, and who's just been handed a forty-word instruction at conversational speed with a Queens accent. That pilot is going to ask for a repeat. And Kennedy Steve is going to give them the repeat, but the tone is going to change. Because every repeat is a tax on the system. Every repeat means he's not talking to someone else who needs an instruction. And the queue is building.
There's a clip — I think it's the one Daniel's probably thinking of — where Kennedy Steve is trying to get a cargo pilot to the cargo ramp. The instruction involved seven turns, three hold-short points, and a crossing of an active runway. The pilot keeps getting lost. Steve's voice goes from controlled irritation to full New York exasperation. And it's hilarious,. But what you're watching is a system failure, not a personality failure.
That's the cargo area incident, and it's the perfect case study. The pilot wasn't incompetent — he was facing a navigation task that would be challenging in a car, let alone in a wide-body aircraft on an unfamiliar airport at night. JFK's cargo area is on the north side of the field, and getting there from the passenger terminals requires navigating through some of the most complex intersections on the airport. You're crossing active runways, you're passing through multiple ground control sectors, you're switching frequencies. And the chart you're looking at is a static diagram — it doesn't show you where you are in real time unless you've got a moving map, and even then, the GPS accuracy on the ground isn't always precise enough to distinguish between two parallel taxiways that are two hundred feet apart.
The moving map helps, but it doesn't solve the problem.
It helps until it doesn't. And this brings us to the technology gap, which is where things get really interesting. While en-route air traffic control uses advanced automation — ERAM, the En Route Automation Modernization system, ADS-B surveillance, digital datalink — ground control at most major US airports still relies on what's called progressive taxi. Which is a fancy way of saying the controller paints lines on a radar screen and hopes pilots follow them. There is a surface surveillance system called ASDE-X that exists at JFK and about thirty-five other US airports. It uses radar and multilateration to track aircraft and vehicles on the ground. But it's underutilized. Controllers are still issuing verbal instructions, pilots are still reading them back, and the whole thing runs on voice and trust.
ASDE-X — that's the system that shows the controller where everyone is on the ground, right? Little blips moving around the airport diagram?
Yes, and it's been around for years, but it's essentially a surveillance tool, not an automation tool. It shows you where things are. It doesn't route them. It doesn't deconflict. It doesn't generate instructions. The controller is still doing all of that manually, in their head, while talking to twenty aircraft simultaneously. Compare this to Tokyo Haneda, which handles similar traffic density to JFK but with roughly half the ground incidents. Haneda uses automated taxi routing and digital datalink for ground instructions. Pilots receive their taxi clearance as a text message on a screen in the cockpit. They read it, they follow it, they don't have to interpret a forty-word verbal transmission through a heavy accent. The error rate plummets.
JFK is running a ground control system that's fundamentally analog in an increasingly digital world. But I have to ask — why hasn't this been fixed? If Haneda has demonstrated that digital datalink works, and the FAA has known about this for years, what's the holdup?
Money and fragmentation. The FAA's budget for surface automation gets spread across hundreds of airports, and JFK — despite being one of the busiest in the world — is just one line item. The Port Authority of New York and New Jersey operates the airport, the FAA provides the air traffic control, and the airlines pay landing fees. When it comes time to invest in surface automation, everyone points at everyone else. The FAA says the Port Authority should fund the infrastructure. The Port Authority says it's an FAA operational responsibility. The airlines say they're already paying enough in fees. And so nothing happens. Meanwhile, Haneda is a single, integrated operation where the airport authority, the navigation service provider, and the major airline are all essentially working from the same budget and the same strategic plan. They can make a decision and implement it without three layers of inter-agency negotiation.
It's not a technology problem. It's a governance problem.
It's almost always a governance problem. The technology exists. It's been demonstrated. The safety case is clear. But the institutional will to deploy it at scale isn't there, because the costs are concentrated and the benefits are diffuse. An airline that operates out of JFK might save millions in fuel and delay costs over a decade if surface automation were implemented, but they'd have to pay for it now, and the savings would accrue over years, and their quarterly earnings report doesn't care about a ten-year ROI.
The stress you hear in Kennedy Steve's voice is the sound of that institutional failure being downloaded onto one person with a radio.
And the stress you hear in Kennedy Steve's voice is the sound of that analog system hitting its limits. He's juggling twenty aircraft in a space smaller than a shopping mall, with nothing but a radio and a radar screen, and every pilot who hesitates, every readback that's wrong, every aircraft that stops in the wrong place — it cascades. A fifteen-minute ground hold at JFK doesn't just delay that one flight. It means missed slots at Heathrow, at Frankfurt, at Dubai. The ripple effects are global.
This is the knock-on effect that I think Daniel was circling around. The stress isn't just about the individual moment — it's about the systemic fragility that Kennedy Steve is essentially the canary in the coal mine for.
There was a ground stop at JFK in twenty twenty-three — a single taxiway closure, just one, due to pavement work — that cascaded into a four-hour gridlock and stranded over ten thousand passengers. The entire system seized up because there's no slack in the ground movement network. Every aircraft has a specific route, a specific gate, a specific departure slot, and when one piece of that puzzle gets removed, the whole thing backs up like a clogged drain. And the ground controller is the person standing there with the plunger.
That's a vivid image.
It's an accurate one. And this is where we get to the core paradox that Daniel identified. Why is ground control harder than air control? It's because the ground is the only phase of flight where aircraft are simultaneously high-density, low-altitude, and human-driven. The air has rules and separation standards and automation. The ground has paint and hope.
Paint and hope. That's the unofficial motto of JFK ground control.
It really is. And the human factor compounds everything. Kennedy Steve's style — the rapid-fire instructions, the New York impatience, the zero-tolerance for confusion — it works because he's good at it. He's been doing it for decades. He knows the airport diagram like you know your living room. But the system is designed for the lowest common denominator pilot. A first officer on their first trip to JFK, flying in from a country where English is a second or third language, arriving after an eight-hour transatlantic flight, at night, in weather. That pilot is going to struggle. And the system doesn't accommodate struggle gracefully.
That's where the viral clips come from. It's not that the pilots are incompetent — it's that they're being asked to perform at the limit of human cognitive capacity in a second language, and Kennedy Steve is the human manifestation of a system that has no slack.
There's a linguistics paper in this somewhere. Taxi instructions are the most complex speech act in aviation because they combine spatial navigation, procedural compliance, and time pressure into a single transmission. You're not just being told where to go — you're being told where to go, where to stop, what to cross, what frequency to switch to, and what to expect next. All in one breath. And you have to read it back correctly. If you're a native English speaker, this is hard. If you're not, it's borderline impossible.
Which makes me wonder — are foreign pilots actually the problem, or are they just the most visible symptom of a system that's fundamentally poorly designed?
They're absolutely the symptom, not the cause. The cause is that we're using voice-only communication for the most complex phase of flight, at the busiest airports in the world, with no meaningful automation support. Haneda proves this is solvable. Digital datalink for taxi instructions exists. Automated routing exists. The FAA's NextGen program has been promising surface automation for twenty years. But implementation has been slow, expensive, and politically complicated. So we're still relying on Kennedy Steve and his radio.
What does the future actually look like? Is this a problem that AI eventually solves, or is there something irreplaceable about the human controller?
This is the open question. Companies like Indra and Saab are testing what they call virtual tow systems — essentially automated ground movement where aircraft are routed by algorithm and pilots follow digital instructions on a screen. No voice, no readback, no ambiguity. The technology exists. The challenge is integration, certification, and the fact that airports are chaotic environments with vehicles, wildlife, weather, and unpredictable human behavior. An AI can route twenty aircraft through a taxiway network efficiently. Can it handle a fuel truck that's parked in the wrong spot, or a catering vehicle that's broken down on a taxiway, or a pilot who's just had a medical emergency? Those edge cases are where human controllers earn their pay.
We're not talking about replacing Kennedy Steve with a server rack. We're talking about giving him better tools so that the routine stuff — the forty-word taxi instructions — gets handled digitally, and his cognitive bandwidth is freed up for the exceptions.
That's the vision. And it's not science fiction. It's operational at Haneda. It's being tested at Schiphol. Singapore Changi is moving in this direction. The US is behind, but not because the technology isn't there. It's because the institutional architecture makes adoption slow. But the pressure is building. Every ground delay, every incursion, every viral clip of a stressed controller — it all adds to the case for change.
Kennedy Steve's job isn't going away tomorrow. But the stress might be.
If we're smart about it, yes. The next frontier in aviation automation isn't pilotless planes — it's automated ground movement. And the payoff is enormous. Fewer delays, fewer incidents, less stress on controllers, less confusion for pilots. But it requires investment, and airports are notoriously bad at investing in infrastructure that passengers don't see.
Because you can't put a plaque on a digital datalink system and charge higher landing fees for it.
It's invisible. Unlike a new terminal with a food court and a yoga room, surface automation doesn't photograph well. But it's what actually makes the airport work. Or in JFK's case, what would make it work better than it does.
For anyone listening who's watched those Kennedy Steve compilations and laughed — and they are funny — what should they take away from this?
First, for the aviation enthusiasts: Kennedy Steve isn't a jerk. He's a symptom of a system that demands perfection from humans in an environment that allows zero margin. His impatience is functional. Second, for the tech-minded: the real automation revolution in aviation isn't in the cockpit — it's on the ground. That's where the biggest gains in safety and efficiency are waiting to be made. And third, for everyone who's ever sat on a plane waiting for a gate and wondered what the holdup is — now you know. Someone is juggling twenty aircraft in a space smaller than a shopping mall, with nothing but a radio and a radar screen, and every single one of those aircraft has to be in exactly the right place at exactly the right time. The fact that it works at all is remarkable.
The fact that it works at all. That's the real takeaway. Next time you hear a Kennedy Steve clip, don't laugh at the pilot — marvel that the system functions.
Maybe write your congressman about funding surface automation at major airports.
There's the civic activism side hustle coming through.
I contain multitudes.
To Daniel's original question — why is Kennedy Steve so stressed, and why is ground control paradoxically harder than air control — the answer turns out to be a combination of geometry, language, technology gaps, and human factors. The ground is where density meets manual control meets voice-only communication. It's the aviation equivalent of a four-way stop with no traffic lights, except everyone's driving a hundred-and-fifty-ton vehicle and half the drivers don't speak the local language.
The geometry of JFK specifically makes it one of the hardest ground control jobs in the world. Four runways, over thirty taxiways, more than a hundred intersections. It's not an airport — it's a puzzle box that someone decided to land airplanes on.
Puzzle box with jet bridges. That's the episode.
One last thought, though. There's something almost admirable about Kennedy Steve's approach. He's not polite, but he's honest. He doesn't sugarcoat. He expects competence because the system requires competence, and when it's not there, he lets you know. In an industry that's increasingly automated and sanitized and scripted, there's something refreshing about a guy who just says what he means at the speed of thought.
The human element, for better and worse. Which is probably why those compilations have millions of views. People recognize real stress, real competence, real friction. It's not ASMR — it's anti-ASMR. And it's compelling.
That's Kennedy Steve's genre. Someone should put that on a t-shirt.
I'm sure someone already has. And now: Hilbert's daily fun fact.
Hilbert: The kabaddi sub-variant known as Sanjeevani kabaddi, where a single revived player can re-enter the match after being tagged out, survived as a competitive format solely through a rulebook printed in nineteen twelve by a British colonial sports administrator stationed in Calcutta — every other documented version of the revival rule was lost when the Punjab Kabaddi Association's archives were destroyed by fire in nineteen forty-seven.
...right.
Kabaddi sub-variants. We contain multitudes, but Hilbert contains more.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you enjoyed this, leave us a review wherever you get your podcasts — it helps. We'll be back next week with whatever Daniel sends us.
Until then, be patient with your ground controllers.
